t-110.5121 mobile cloud computing summary 28.11 · • typical public cloud sla promise • 99.95%...
TRANSCRIPT
11/28/2012
Teacher: Yrjö Raivio
Assistant: Eetu Jalonen
Aalto University, School of Science
Department of Computer Science and Engineering
Data Communications Software
Email: yrjo.raivio(at)aalto.fi
Course email: t-110.5121(at)tkk.fi
© Y Raivio
T-110.5121 Mobile Cloud Computing
Summary
28.11.2012
© Y Raivio
Social Networks and Operators
11/28/2012 2
Source: http://techcrunch.com/2012/11/21/the-facebook-mobile-plot-thickens-orange-
inks-deal-for-social-calling-service-starting-summer-2013/
© Y Raivio
• “You have a clear view of the advantages of distributed computing technologies, such as cloud computing, in the mobile space. You understand the core properties as well as the strengths and limitations of mobile cloud computing. You have a vision how clouds will change the mobile ecosystem, reviewed both from technology and business angles.”
• (Mobile) Disruption
• Neutral, scientific and critical view, over the hype
• Technology, Business, Theoretical and Practical approach
Targets
11/28/2012 3
© Y Raivio
• 5 ECTS: 24 + 0 (2 + 0), not applicable to post graduate studies
• Lectures 24 h, Lecture preparation 24 h, Assignments 48 h, Exam preparation
36 h, Exam 3 h
• Lectures are not obligatory but highly recommended
• Exam
• Tue 18.12.2012, 9-12, Tue 08.01.2013, 9-12 or Tue 28.05.2013, 9-12
(remember to register in Oodi)
• Structure:
• 1 obligatory question: 6 definitions
• 3 questions, 2 must be answered
• 1 obligatory essay
• 6 points from each: 0-11=0, 12-13=1, 14-15=2, 16-18=3, 19-21=4, 22-24=5
• Assignments in pairs (possible alone with good reasons)
• Course feedback!!!
Requirements
11/28/2012 4
© Y Raivio
• Exam 50% + Both assignments together 50%
• Both assignments have the same weight, e.g. 25%
• Exam and assignments evaluated 0-5
• To pass the whole course, each component must be passed at least with grade 1
• Example:
• Exam: 5
• Assignment 1: 3
• Assignment 2: 5
• Total: 50% x 5 + 25% x 3 + 25% x 5 = 4.5 = grade 5 (rounded to
closest integer)
Grading
11/28/2012 5
© Y Raivio
1. Armbrust, Michael, Fox, Armando, Griffith, Rean, Joseph, Anthony D., Above the Clouds: A Berkeley View of Cloud Computing, Feb. 10, 2009.
Available at: http://www.eecs.berkeley.edu/Pubs/TechRpts/2009/EECS-2009-28.pdf
2. Lee Badger, Tim Grance, Robert Patt-Corner and Jeff Voas: Draft Cloud Computing Synopsis and Recommendations, Recommendations of the
National Institute of Standards and Technology, May 2011, available at: http://csrc.nist.gov/publications/drafts/800-146/Draft-NIST-SP800-146.pdf
3. Rolf Harms and Michael Yamartino: The Economics of the Cloud, Nov. 2010, available at:
http://www.microsoft.com/presspass/presskits/cloud/docs/The-Economics-of-the-Cloud.pdf
4. Greenberg, Albert, Hamilton, James, Maltz, David A., Patel, Parveen (2009) The cost of a cloud: research problems in data center networks,
SIGCOMM Comput. Commun. Rev., Vol. 39, No. 1, pp. 68-73.
5. Mohammad Hajjat, Xin Sun, Yu-Wei Eric Sung, David Maltz, Sanjay Rao, Kunwadee Sripanidkulchai, and Mohit Tawarmalani, Cloudward Bound:
Planning for Beneficial Migration of Enterprise Applications to the Cloud, ACM SIGCOMM'10, (Sections 1-2, 2 pages)
6. Ross Anderson: Security Engineering, 1st Edition (2001), Chapter 1, available from http://www.cl.cam.ac.uk/~rja14/Papers/SE-01.pdf (12 pages);
Cloud Security Alliance: Top Threats to Cloud Computing V1.0, March 2010, (14 pages), available from
https://cloudsecurityalliance.org/topthreats/csathreats.v1.0.pdf; Security Guidance for Critical Areas of Focus in Cloud Computing, v. 2.1, Dec 2009,
Section II/Domain 5, (6 pages), available from https://cloudsecurityalliance.org/csaguide.pdf
7. Lutz Schaubert and Keith Jeffery: Research in Future Cloud Computing, May 2012, Section II/Domain 5, (6 pages), available at:
http://cordis.europa.eu/fp7/ict/ssai/docs/future-cc-2may-finalreport-experts.pdf (NOT REQUIRED)
8. Antero Juntunen, Eetu Jalonen and Sakari Luukkainen: HTML5 in Mobile Devices: to be published in HICSS-46 in Jan 2013. (10 pages)
9. Matti Kemppainen: Mobile Computation Offloading: A Context Driven Approach, IWORK paper from 2011. (10 pages)
Exam reading material
11/28/2012 6
© Y Raivio
12.09 Introduction, Basics & Assignment 1, Yrjö Raivio and Eetu Jalonen
19.09 Mobile Networks, Jukka K. Nurminen
26.09 Programming on open APIs, Olli Rinne/Apps4Finland
03.10 Mobile Cloud Business, Antero Juntunen
10.10 Scalable Cloud Computing, Keijo Heljanko
17.10 Mobile Cloud, Yrjö Raivio and HTML5, Eetu Jalonen
24.10 No lecture, exam week
31.10 Green Cloud Computing, Tommi Mäkelä & Assignment 2, Yrjö Raivio
07.11 Cloud Computing in Data Centres, Jarno Laitinen, Risto Laurikainen and Peter Jenkins/CSC
16.11 Cloud Security, Abu Shohel Ahmed/Ericsson
21.11 Industry keynote, Markku Lepistö/NSN
28.11 Resource Provisioning & Summary, Yrjö Raivio and Eetu Jalonen
05.12 Spare
Lecture schedule
11/28/2012 7
© Y Raivio
Massive data volumes
11/28/2012 8
Source: Ambrust et al, Above the Clouds: A Berkeley View of Cloud Computing, 2009
Bottleneck: Bandwidth
Example: Facebook 1 PB totally, 2-3 TB added each day
1 TB drive, 1 Gbit/s I/O = 2 h 13 min
T = 1012
P = 1015
© Y Raivio
11/28/2012 9
Some numbers
• Global ICT business size
• 2008: $ 383 B, 4% cloud
• 2012: $494 B, 9% cloud
• Largest growth in storage
• SMEs have best opportunities to adapt
• Also small countries with good infrastructure
• Startup costs for SMEs dropped dramatically
Source: F. Etro, The Economic Impact of
Cloud Computing on Business Creation,
Employment and Output in Europe, 2009
Source: M. Suster, It’s Morning in Venture Capital,
blog, May 23, 2012
© Y Raivio
Mobile internet access doubled in a year
11/28/2012 10
Source: http://mobithinking.com/mobile-marketing-tools/latest-mobile-stats/b#mobilepageviews
© Y Raivio
11/28/2012 11
Mobile capabilities are improving but
battery capacity is still a bottleneck
• Less new services
• More frequent charging
• Physically larger battery
• More energy efficient chips
• Intelligent methods
• Radical battery inventions
• HTML5
Energy
consumption
2000 2005 2010 2015 2020
Basic services (voice,SMS)
New
services
Navigation
Multimedia
Social media Web
Battery
capacity
3D NOK
NOK
NOK
OK
OK
?
Source: professor Jukka K. Nurminen
?
© Y Raivio
(Mobile) Cloud Computing
11/28/2012 12
“Mobile Cloud computing is a model for enabling convenient, on-demand mobile network access to a shared pool of configurable mobile computing resources (e.g., networks, servers, storage, applications, and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction.”
Adapted from: P. Mell and T. Grance, The NIST Definition of Cloud Computing, 2009
© Y Raivio
11/28/2012 13
Definition 2
1. The illusion of infinite computing resources available on demand, thereby eliminating the need for Cloud Computing users to plan far ahead for provisioning.
2. The elimination of an up-front commitment by Cloud users, thereby allowing companies to start small and increase hardware resources only when there is an increase in their needs.
3. The ability to pay for use of computing resources on a short-term basis as needed (e.g., processors by the hour and storage by the day) and release them as needed, thereby rewarding conservation by letting machines and storage go when they are no longer useful.
Source: Ambrust et al, Above the Clouds: A Berkeley View of Cloud Computing, 2009
© Y Raivio
• Cloud performance
• Processing time at the data center
• Processing time on the device
• Network latency
• Data transport time
• Mobile challenges
• Variable reliability, less throughput, longer latency
• Energy
• Limited resources, example Dell Desktop vs. iPhone 4 • 3 times less processing power
• 8 times less memory
• 5 times less storage capacity
• 10 times less network bandwidth
• But mobiles are always with you and provide context data (location, sensor data, camera)
Fixed vs. Mobile
11/28/2012 14
Source: http://www2.alcatel-lucent.com/blogs/techzine/2010/mobile-cloud-computing-challenges/
© Y Raivio
• Mobile Cloud Computing: mobile apps are processed and data stored in a cloud rather than on the native handset, referred as Mobile Computing
• Mobile Offloading: moving data, code block or virtual machine from mobile to cloud or vv.
• Using mobile context to enhance cloud based services
• Cloud within mobile (Hyrax, Cuckoo, ThinkAir, MAUI, CloneCloud..)
What is Mobile Cloud Computing (MCC)
11/28/2012 15
Cloud
End users
Mobile
Offloading
© Y Raivio
11/28/2012 16
• Deployment models (Public, Private, Hybrid, Community Cloud)
• Service models (SaaS, PaaS, IaaS)
• Key Benefits
• Economies of scale
• Elasticity
• Resource planning
• Pay-as-you-go
• Always available
• Technology
• Virtualization, Storage, SLA, Provisioning, Energy, Security, Mobility
Key issues
Source: Armbrust, Michael, Fox, Armando, Griffith, Rean, Joseph, Anthony D., ”Above the Clouds:
A Berkeley View of Cloud Computing”, Feb. 10, 2009.
© Y Raivio
11/28/2012 17
• Public cloud
• Resources made available to the general public via the Internet
• Scalable
• Pay for what you use
• Private cloud
• Host own resources
• Provide to internal customers only
• Provision with cloud interfaces
• Hybrid cloud
• Some resources provided internally and some outsourced
• Community cloud
• Address special needs of a community
Deployment models
© Y Raivio
11/28/2012 18
Main benefits - Service models
© Y Raivio
11/28/2012 19
Economies of scale
Source: Rolf Harms and Michael Yamartino: The Economics of the Cloud, Nov. 2010.
• Cheaper MIPS (5-7 times)
• Better utilization of computing resources (5-10% to 60-80%)
• Multi-tenancy: one instance can serve several customers
• Less admin people per server (from 1:100 up to 1:10 000)
• Worth 1$ IT requires 8$ admin costs
© Y Raivio
11/28/2012 20
Statistical multiplexing
0
10
20
30
40
50
60
70
80
90
100
0 6 12 18 24
Lo
ad
Time
Statistical multiplexing across time zones
Country1
Country2
Country3
Country4
Average
© Y Raivio
11/28/2012 21
Elasticity – pay-as-you-go
Source: Ambrust et al, Above the Clouds: A Berkeley View of Cloud Computing, Feb 2009
Overprovisioning Underprovisioning
Lost customers due to underprovisioning • Avoid high upfront investment, avoid risk
• Adapt to changing business
• Buy or lease
• Amortizise value to investment period
© Y Raivio
11/28/2012 22
Amortization
• Hybrid cloud
• Public vs. Private
• www.cloudonomics.com
• Cost structure by Greenberg (2009):
• ∼45% Servers CPU, memory, storage
• ∼25% Infrastructure
• ∼15% Power draw
• ∼15% Network Links
• Staff? 1$ IT : 8$ Admin!
• Equated Monthly Installment
• Net Present Value
1)1(
)1(
12
1212
mr
mrr
m EA
Er
S
r
CPNPV
N
N
T
TT
)1()1(0
Source: TeliaSonera
Case SMSC
E = basic investment
N, m = length of the investment
R = rate of interest
Pt = annual revenue
Ct = annual cost
S = residual value
© Y Raivio
11/28/2012 23
Always available
• Anyone, anytime, anywhere
• High availability?
• Typical public cloud SLA promise
• 99.95% = max 4 h 23 min down time per year
• Telecom
• 99.999% = 5 min
• Availability Zone, fully (?) independent computing systems
• Using two Availability Zones
%9999.99)1(11 22
APP FP
© Y Raivio
11/28/2012 24
Animoto:
In 3 days
from 50 to
3500 servers
Resource planning
• Resources can be optimized to meet service needs
• Service integration time can be shortened, example Short Message Service Center setup from 2 weeks to 4 minutes
Time
Turnover
LAMP Public
cloud
Private
or hybrid
Cloud,
Hosted
Solution...
Early
development
Exponential
growth
Mature
market
Source: Rolf Harms and Michael Yamartino: The Economics
of the Cloud, Nov. 2010.
© Y Raivio
• Mobile communication (19.9, 21.11)
• Virtualization, storage and programming models (10.10)
• Mobile offloading and HTML5 (17.10)
• Energy (31.10)
• Data Centres (7.11)
• Security (14.11)
• Resource provisioning (28.11)
Key technology issues
11/28/2012 25
© Y Raivio
11/28/2012 26
Source: Kumar & Lu, ”Cloud Computing for Mobile Users:
Can Offloading Computation Save Energy ”, 2010
Source: Chun and Maniatis, ”Augmented Smartphone
Applications Through Clone Cloud Execution”, 2009
Mobile offloading
© Y Raivio
11/28/2012 27
Virtualization
1) Cost Savings
2) Better Resource Utilization
3) Better Memory Management
4) Increased Availability
5) Better Resource Provisioning
6) Energy Saving
APP
1
OS OS
CPU CPU
APP
N
CPU
2
Source: Z. Ou, Virtualization Technology, T-110.7100, Autumn 2010.
..
OS OS
CPU CPU
APP APP APP
1
OS OS
APP
N ..
CPU
1
Hypervisor Hypervisor
Virtual Machines Virtual Machines
..
Single task Multi task Hyper threading Virtualization
APP
1
OS-1 OS-1
APP
N .. APP
1
OS-N OS-N
APP
N ..
© Y Raivio
• Freely available data from government, science, business and other organizations, citizens…
– free access and absence of technological restrictions
– License allows redistribute and reuse
– License may require attribution and integrity
• Open data promotes
– Democracy and transparency
– New and improved processes (and easier live)
– New business opportunities
– Maker – Do-It-Yourself– culture
More info from Open Knowledge Foundation: http://okfn.org/
Open Data and APIs
11/28/2012 28
© Y Raivio
• A collection of technologies aimed to provide elastic “pay as you go” computing
• Virtualization of computing resources: Amazon EC2, Eucalyptus, OpenNebula, Open Stack Compute, . . .
• Scalable file storage: Amazon S3, GFS, HDFS, . . .
• Scalable batch processing: Google MapReduce, Apache Hadoop, PACT, Microsoft Dryad, Google Pregel, Spark,..
• Scalable datastore: Amazon Dynamo, Apache Cassandra, Google Bigtable, HBase,. . .
• Distributed Coordination: Google Chubby, Apache Zookeeper, . . .
• Scalable Web applications hosting: Google App Engine, Microsoft Azure, Heroku, . . .
Cloud computing technologies
11/28/2012 29
© Y Raivio
• In a PODC 2000 conference invited talk Eric Brewer made a conjecture that it is impossible to create a distributed asynchronous system that is at the same time satisfies all three CAP properties:
• Consistency
• Availability
• Partition tolerance
• This conjecture was proved to be a Theorem in the paper: “Seth Gilbert and Nancy A. Lynch. Brewer’s conjecture and the feasibility of consistent, available, partition-tolerant web services. SIGACT News, 33(2):51-59, 2002.”
• However, Google published Spanner concept based on atom clock 10/2012…
Brewer’s CAP theorem
11/28/2012 30
© Y Raivio
11/28/2012 31
• Power Usage Efficiency (PUE) defined as:
• Typical PUE 2-3, state of the art 1.1
• 15% of all costs
• 59% IT equipment
• 33% cooling
• 8% power loss
• Power off idle machines
• Raise temperature
• Cold location, cheap energy (for example Finland)
• More fine-grain accounting
• Better algorithms
• E2E model required including public clouds
tPowerITEquipmen
ityPowerTotalFacilPUE
Source: A. Greenberg, J. Hamilton, D.A. Maltz abd P. Patel, The Cost of a Cloud: Research
Problems in Data Center Networks, ACM SIGCOMM Computer Comm. Review, Jan 2009.
Power management
© Y Raivio
11/28/2012 32
• Top threats
• Abuse and Nefarious Use of Cloud Computing
• Insecure Application Programming Interfaces
• Malicious Insiders
• Shared Technology Vulnerabilities
• Data Loss/Leakage
• Account, Service & Traffic Hijacking
• Unknown Risk Profile
Security
Source: Cloud Security Alliance, “Top Threats to Cloud Computing V1.0”, March 2010
© Y Raivio
11/28/2012 33
Pros and cons
•Remote and shared computing over the Internet
•Consists of components that communicate through APIs
!
•Simple architecture
•Efficient usage of CPU (>50%)
•Scalability
•Load balancing
•Low capex
•High availability
?
•Security & Privacy
•High usage of certain CPUs
•Interoperability
•Vendor lock-in
•High opex
•SLA critical
© Y Raivio
11/28/2012 34
Conclusions
• Cloud computing is a new business model
• Great tool to startups
• Biggest challenge: (lack of) trust
• Next target: utility computing (similarly to water, electricity, gas and telephony)
• Future research topics
• Energy efficiency in data centers
• Cloud interoperability
• Security
• HTML5
• Dynamic resource provisioning algorithms
• Game theory
© Y Raivio 11/28/2012
Questions?
Contacts:
Teacher: Yrjo Raivio, A122
Assistant: Eetu Jalonen A121
Course staff: t-110.5121(at)tkk.fi